An Educators' Guide to Schoolwide Reform

An Educators' Guide to Schoolwide Reform

APPENDIX C
SUMMARY OF FINDINGS ON
STUDENT ACHIEVEMENT EFFECTS

Summary of Effects on Student Achievement for
Accelerated Schools

Source	Research Rating	Measurement Instrument	Group/Subgroup	Dimension Measured	Effect	Test Statistic
Knight, Stallings (1995)	3.1	Metropolitan Achievement Test	Accelerated School (AS) vs. matched control school before implementation (1989), after implementation (1991)	Reading	Achievement in AS rose from below grade level to above grade level and above achievement in comparison school for 1^st, 2^nd, 3^rd and 5^th but not 4^th grades.	Mean grade equivalents
		Metropolitan Achievement Test		Language	Achievement in AS began at or above grade level and near achievement for control schools; rose above grade level and control school for 1^st, 2^nd, and 3^rd grades.	Mean grade equivalents
		Texas Assessment of Academic Skills	Accelerated School (AS) vs. matched control school, district average, state average	Math	AS had higher percent of students passing in 3^rd grade, but lower percent passing in 5^th grade, compared to all 3 control groups.	Percent passing
				Reading	AS had higher percent of students passing in 3^rd grade, but lower percent passing in 5^th grade, compared to all 3 control groups.	Percent passing
				Writing	AS had higher percent of students passing in 3^rd grade and slightly higher or lower percent of students passing in 5^th grade compared to control groups.	Percent passing
Ross, Alberg, McNelis (1997)	3.3	Durrell Oral Reading; Woodcock Passage Comprehension; Woodcock Word Attack, Writing Sample	1 Accelerated School (AS), 1 Success for All (SFA), and 2 locally developed programs (LDP); 43%-54% free/reduced lunch	Reading/Writing	LDP schools performed highest of the 3 on Oral Reading (ES=+.49) but lowest on Word Attack (ES=-.36). AS performed highest on Word Attack, but lowest on Writing (ES=-.30). SFA scored highest on Writing (ES=+.25), but lowest on Passage Comprehension (ES=-.49).	Mean score; effect size
Ross, Alberg, McNelis (1997)	3.3		3 AS and 3 SFA schools; 62%-71% free/reduced lunch	Reading/Writing	SFA outperformed AS on Word Attack (ES=.28), but did significantly worse than AS on Writing (ES=-.49).	Mean score; effect size

Summary of Effects on Student Achievement for America’s Choice

No studies on effects of America’s Choice on students are publicly
available. However, the developer collects and makes publicly available
student test data from America’s Choice schools.

Summary of Effects on Student Achievement for
ATLAS Communities

No studies on effects of ATLAS Communities on students are publicly
available. According to the developers, most of the research on ATLAS
to date has been formative, and has led to improvements in the approach.
Further, the developers, New American Schools, and participating
pathways collect and make publicly available test data on student effects.

Summary of Effects on Student Achievement for Audrey Cohen
College: Purpose-Centered Education

No studies on effects of Audrey Cohen College: Purpose-Centered Education
on students are publicly available. However, the developers collect and make
publicly available standardized test score data on Audrey Cohen College:
Purpose-Centered Education.

Summary of Effects on Student Achievement for the
Basic Schools Network

No studies on the effects of the Basic Schools Network are publicly available.
However, the developer collects and makes publicly available test data, and
will release a full report on the Basic Schools Network in spring 1999.

Summary of Effects on Student Achievement for Coalition
of Essential Schools

Source	Research Rating	Measurement Instrument	Group/Subgroup	Dimension Measured	Effect	Test Statistic
McCarthy (1994)	3.0	Delaware Educational Assessment Program	11^th-grade students	Math, Verbal composites	Schoolwide average scores improved from 1989 to 1992.	na
Stringfield et al. (1997)	3.1	CTBS	Students in 4 CES schools	Reading Math	Test scores dropped over time. Test scores dropped over time.	na na

Summary of Effects on Student Achievement for Community for
Learning

Source	Research Rating	Measurement Instrument	Group/Subgroup	Dimension Measured	Effect	Test Statistic
Brookhart, Casile, McCown (1997)	3.1	ITBS	Community for Learning Students	Math Achievement	Students with 0, 1, 2, or 3 years of instruction via CFL did not show differences in ITBS scores. However, the author asserts that this may be due to a ceiling effect.	Analysis of covariance
		Student folders (incl. Student Achievement Summary and Student Annotation Forms)		Student Achievement	Teacher ratings of mastery of objectives were similar across 3 school years. Percent achievement of objectives was rated as 83.3% on average, with a large amount of variability. Percent achievement in 1995/1996 was higher than in 1994/1995, but similar to 1993/1994.	na
Oates, Flores, Weishew (1997)	3.1	District-wide standardized test	House using CFL vs. test at school	Reading, math	CFL scores slightly (not significantly) higher; CFL sample included special education and bilingual students, control sample did not.	ANOVA
		Enrollment rates	House using CFL vs. rest of school (long follow-up)	Enrollment	CFL – 81% regular, 100% special education students still enrolled; control – 40% regular, 52% special education students still enrolled.
		Grade level	House using CFL vs. rest of school (long follow-up)	Grade level	CFL – 50% regular, 43% special education students on grade level; controls – 26% regular, 17% special education students on grade level.
Reynolds, Heistad, Peterson, Dehli (1992)	3.4	Time to complete units of instruction	Non-labeled, Chapter 1, and special education CFL elementary students	Pace of learning for math and reading	Although all students can master the same material, regular education students move through the curriculum the fastest, followed by Chapter 1 students, and then Special Education students. Differences evident from first grade. Placement in a categorical program is related to slower learning.	t-test

Summary of Effects on Student Achievement for Community
for Learning (Continued)

Source	Research Rating	Measurement Instrument	Group/Subgroup	Dimension Measured	Effect	Test Statistic
Wang, Oates, Wieshew (1995)	3.1	Districtwide standardized test; NCEIC survey	School 1: CFL school vs. two comparable local schools (1992-1994)	Reading	CFL reading scores rose over 2 years; control school scores went down.	na
				Math	CFL math scores rose over 2 years; control school scores went down.
				Perceptions	CFL students had significantly higher perceptions of feedback, aspirations for self, self-concept, task orientation, and rules clarity compared to non-CFL students.	F-test
			School 2: CFL classes vs. other classes, same school	Reading	CFL students had higher scores (41) than non-CFL students (32).
				Math	CFL students had higher scores (46) than non-CFL students (34).
				Perceptions	CFL students had significantly higher perceptions of satisfaction, pacing, teacher aspirations for students, involvement, affiliation, teacher support, and order compared to non-CFL students.	F-test
			School 3: CFL classes vs. other classes, same school	Reading, Math	CFL students scored slightly (not significantly) higher than non-CFL students.
			School 3: CFL classes vs. other classes, same school	Perceptions	CFL students had significantly higher perceptions of a multicultural, social, active, non-traditional, guided interdisciplinary instructional environment, and sensed greater affiliation, teacher support, and participation compared to non-CFL students.	F-test

Summary of Effects on Student Achievement for Community
for Learning (Continued)

Source	Research Rating	Measurement Instrument	Group/Subgroup	Dimension Measured	Effect	Test Statistic
Wang, Peverly, Randolph (1984)	3.1	Stanford Diagnostic Mathematics Test; California Achievement Test in Reading	Students in classes with varying levels of CFL implementation	Student achievement in Reading and Math	Achievement gains for regular education students were significantly greater than the national norm. Achievement gains for mainstreamed special education students were significantly greater than expected gains. Mean overall percentiles for special education students were above the national norm. For both math and reading, implementation scores account for a significant proportion of the variance in achievement. The overall correlation of reading gains and implementation was not significant, however, and not all subscales of implementation significantly correlated with achievement gains (e.g., establishing/communicating rules and developing student self-responsibility).	Percentiles, Regression
		Intellectual Achievement Responsibility Scale (IAR)		Achievement responsibility	Overall significant positive correlation between total IAR score and total implementation score.	Correlations, Regression

Summary of Effects on Student Achievement for Co-NECT

No studies on effects of Co-NECT on students are sufficiently rigorous
to report the findings here. However, the developer collects and makes
publicly available student test data.

Summary of Effects on Student Achievement for Core Knowledge

Source	Research Rating		Measurement Instrument	Group/Subgroup	Dimension Measured	Effect	Test Statistic
Oklahoma City Public Schools (1996)	3.5		Iowa test of Basic Skills	3 Core Knowledge schools vs. matched control schools	Total	Core students had (not significantly) higher scores for 3 years; significantly higher than Great Expectations schools.
					Reading comprehension	Core students had (not significantly) higher scores for 3 years; significantly higher than Great Expectations schools.
					Language	Core students had higher scores for 3 years (significant in one year); significantly higher than Great Expectations schools.
			Writing exercise	3 Core Knowledge schools vs. matched control schools	Writing	Students in core magnet schools performed better than comparisons. Students in core non-magnet schools did not perform as well as comparisons.
Stringfield, McHugh (1996)		3.1	Comprehensive Test of Basic Skills	6 Core Knowledge schools vs. matched control schools
				1^st grade	Reading comprehension	Core students gained more (10.83) than control students (7.0).	Gain scores
				1^st grade	Math concepts	Core students gained more (2.33) than control students (.50).	Gain scores
				3^rd grade	Reading comprehension	Core students gained more (.17) than control students (-1.50).	Gain scores
				3^rd grade	Math concepts	Core students gained less (4.5) than control students (4.67).	Gain scores

Summary of Effects on Student Achievement for Core
Knowledge (Continued)

Source	Research Rating	Measurement Instrument	Group/Subgroup	Dimension Measured	Effect	Test Statistic
Stringfield, McHugh (1998)	3.3	Maryland State Performance Assessment Program	Core Knowledge vs. control; Core Knowledge vs. all MD schools
			3^rd grade	Reading	Core > control; Core > MD schools.	Not reported
				Math	Core > control; Core > MD schools.
				Social studies	Core > control; Core > MD schools.
				Science	Core > control; Core > MD schools.
				Writing	Core > control; Core > MD schools.
				Language	Core > control; Core > MD schools.
				All subjects	Core > control; Core > MD schools.
			5^th grade	Reading	Core > control; Core < MD schools.
				Math	Core > control; Core > MD schools.
				Social studies	Core > control; Core > MD schools.
				Science	Core < control; Core < MD schools.
				Writing	Core > control; Core > MD schools.
				Language	Core > control; Core < MD schools.
				All subjects	Core > control; Core = MD schools.
			6 Core Knowledge Schools vs. 6 control schools, state average	Reading	Core students (7.8%) gained similar to control students, more than state average (3.5%).	Gain scores
				Math	Core students (9.5%) gained more than control students (7.6%) and state average (8.2%).	Gain scores
				Social studies	Core students (9.5%) gained more than control students (7.6%) and state average (5.7%).	Gain scores
				Science	Core students (8.4%) gained less than control students (15.8%) but more than state average (6.4%).	Gain scores
				Writing	Core students (12.2%) gained more than control students (8.9%) and state average (4.2%).	Gain scores
				Language	Core students (13.4%) gained more than control students (10.2%) and state average (8.9%).	Gain scores

Summary of Effects on Student Achievement for Different
Ways of Knowing

Source	Research Rating	Measurement Instrument	Group/Subgroup	Dimension Measured	Effect	Test Statistic
Catterall (1995)	3.3	Standardized test	DWoK students, gains over three years	Math	Student achievement increased slightly (not significantly) for every year of exposure to DwoK.	ANCOVA, F-test
				Language arts	Students gained 8 percentile points (a significant gain) for every year of exposure to DwoK.	ANCOVA, F-test
				Social studies	DwoK students outscored control students by .5 to .75 points (on a three-point scale).	Mean score
		Writing samples	Students with 3 years of DWoK vs. no DWoK	Reading	DwoK students had somewhat higher grades.	Mean grades
		Grades	3 years DWoK vs. no DWoK	Math	DwoK students had significantly higher grades (for one of two samples).	Mean grades
				Language arts	DwoK students had significantly higher grades (both samples).	Mean grades
				Social studies	DwoK students had significantly higher grades (for one of two samples).	Mean grades

Catterall, Dreyfus, DeJarnette (1995)	3.2	California Achievement Test	DWoK students 1^st (1992-93) and 3^rd (1994-95) years of implementation	Reading	Achievement increased from 37.9 to 42.3 percentile.	Percentiles
		California Achievement Test		Math	Achievement increased from 52.7 to 56.5 percentile.	Percentiles
		Writing samples	DWoK vs. non-DWoK students	Social studies	DWoK students scored higher on pre- and post-test than control students. In a comparison of "best of two" scores, DWoK students in grade 3 scored slightly higher than control students; DWoK stu-dents in grade 5 scored significantly higher than control students.	Mean scores

Summary of Effects on Student Achievement for Different
Ways of Knowing (Continued)

Source	Research Rating	Measurement Instrument	Group/Subgroup	Dimension Measured	Effect	Test Statistic
Kentucky Department of Education (1998)	3.1	KIRIS	High vs. low vs. no implementation:
			All schools	Test score index (across subjects)	High and low implementation DWoK schools had slightly (not significantly) higher gain scores than non-DwoK schools.	Gain scores, t-test¹
			Title I schoolwide schools	Test score index (across subjects)	High implementation schools had significantly higher gain scores than non-DWoK schools.	Gain scores, t-test
			Title I targeted assistance schools	Test score index (across subjects)	High and low implementation DWoK schools had slightly (not significantly) higher gain scores than non-DWoK schools.	Gain scores, t-test
			Not Title I schools	Test score index (across subjects)	High and low implementation DWoK schools had slightly (not significantly) lower gain scores than non-DWoK schools.	Gain scores, t-test

________________________

1 T-tests were conducted for this report based on data given in the study.

Summary of Effects on Student Achievement for
Different Ways of Knowing (Continued)

Source	Research Rating	Measurement Instrument	Group/Subgroup	Dimension Measured	Effect	Test Statistic
Petrosko (1997); Hovda, Kyle (1997); Wang, Sogin (1996)²	3.1	KIRIS	Demonstration sites vs. other DWoK sites vs. no DWoK (over two years (1993-95)	Test score index (across subjects)	Demo sites had higher percentage gains (74%) and higher final scores (39.4) than other DWoK sites (61%, 38) or controls (59%, 37.5).	Gain scores
				Reading	Demo sites had higher percentage gains (86%) and higher final scores (56.4) than other DWoK sites (74%, 55.3) or controls (73%, 55).	Gain scores
				Math	Demo sites had higher percentage gains (141%) and higher final scores (42.7) than other DWoK sites (95%, 42.4) or controls (91%, 41.3).	Gain scores
				Science	Demo sites had higher percentage gains (92%) but lower final scores (31.8) than other DWoK sites (82%, 32.6) or controls (82%, 32.5).	Gain scores
				Social studies	Demo sites had higher percentage gains (63%) and higher final scores (40.8) than other DWoK sites (45%, 38.9) or controls (43%, 38.4).	Gain scores
		Survey	Schools having a teacher trained in DWoK 1995-96	Teacher perceptions of effects of DWoK	Over 90% of teachers surveyed felt that DWoK made learning more effective for students.	Percent of responses

________________________

2 These three studies were produced and reviewed as one set.

Summary of Effects on Student Achievement for Direct Instruction

Source	Research Rating	Measurement Instrument	Group/Subgroup	Dimension Measured	Effect	Test Statistic
Adams, Engelmann (1996)	3.8	Meta-analysis	DI vs. comparison	Overall achievement	Effect Size (ES) of 0.97 overall; 0.87 average per study.	Effect size¹
Adams, Engelmann (1996)	3.8	Meta-analysis	DI vs. comparison	Elementary vs. secondary/adults Language Reading Social skills Mathematics Spelling Health Science Type of test Type of research design Duration of intervention Type of teacher Fidelity of implementation Country	Elementary, ES of 0.78 per study; secondary/adult ES of 1.11 per study. ES of 0.49. ES of 0.69. ES of 0.97. ES of 1.11. ES of 1.33. ES of 1.60. ES of 2.44. Norm referenced tests, ES of 0.57; Criterion-referenced test, ES of 1.48. Casual comparative, ES of 1.20; experimental, ES of 0.85. Up to 1 year, ES of 0.95 per study; over 1 year, ES of 0.78 per study. Regular teacher, ES of 0.84 per study; specially trained teacher, ES of 0.92 per study. Fidelity checks in study, ES of 0.96 per study; no fidelity checks in study, ES of 0.74 per study. United States, ES of 0.87 per study; non-United States, ES of 0.85 per study.	Effect size Effect size Effect size Effect size Effect size Effect size Effect size Effect size Effect size Effect size Effect size Effect size Effect size Effect size

Summary of Effects on Student Achievement for Direct Instruction

Source	Research Rating	Measurement Instrument	Group/Subgroup	Dimension Measured	Effect	Test Statistic
Becker, Carnine (1980)	3.4	Metropolitan Achievement Test—Elementary Level, Form F (1970)	DI vs. DI other models of reform at Follow-Through site vs. non-Follow- Through site	Reading	DI scored +214 on ISO, higher than any other model.	Index of significant outcomes (ISO)
				Mathematics	DI scored +429 on concepts and +571 on problem solving, higher than any other model.	ISO
		Coopersmith Self-Esteem Inventory (1967)		Affect (themselves, school)	DI outscored other models (score not available).	ISO
		Intellectual Achievement Responsibility Scale (IARS) (1965)		Extent to which children attribute success or failure to themselves or outside	DI outscored other models (score not available).	ISO
		Metropolitan Achievement Test (MAT)		Reading	DI scored 41^st percentile; next best was Behavior Analysis in 34^th percentile.	Percentile
				Mathematics	DI scored 48^th percentile; next best was Behavior Analysis in 28^th percentile.	Percentile
				Language	DI scored 50^th percentile; next best was Bank Street in 23^rd percentile.	Percentile
Becker, Gersten (1982)	3.7	Wide Range Achievement Test (WRAT)	DI vs. comparison	Reading Reading Mathematics Spelling	Significantly higher achievement effects for DI in grades 5, 6, and for both grades combined. Significantly higher achievement effects for DI in grade 5, not in 6, and combined for both grades. No significant effects for DI in grade 5, significantly higher achievement effects for grade 6 and for both grades combined. Significantly higher achievement effects for DI in grades 5, 6, and for both grades combined.	c² c² c² c²
		MAT		Reading Reading Mathematics Spelling		c² c² c² c²
		MAT		Language	No significant effects for DI in grade 5, significantly higher achievement effects for grade 6 and for both grades combined.	c²

Summary of Effects on Student Achievement for Direct Instruction

Source	Research Rating	Measurement Instrument	Group/Subgroup	Dimension Measured	Effect	Test Statistic
Gersten (1985)	3.6	Narrative literature review	DI vs. comparison	Beginning reading instruction with high risk students	No clear effects	Not provided
				Language,	Significantly higher achievement for the DI students.	Not provided
				instruction for the moderately retarded
				Teaching reading to learning disabled students	Significantly higher achievement for the DI students.	Not provided
				Reading	Significantly higher achievement for the DI students.	Not provided
				Longitudinal growth of moderately retarded children	Moderately retarded children gained at a significantly faster rate than nonhandicapped peers.	Not provided
				Effects of DI on mainstreamed, low IQ students	Significantly higher achievement for the DI students.	Not provided
Gersten, Becker, Heiry, White (1981)	3.6	WRAT and MAT	Low-income children in primary grades divided into 6 IQ blocks; all in DI classes; longitudinal analysis	Reading	No difference in rate of growth of achievement by IQ; that is, those entering with lower IQ, end with lower IQ.	F–test
		WRAT and MAT		Mathematics	No difference in rate of growth of achievement by IQ; that is, those entering with lower IQ, end with lower IQ.	F–test
		Slosson Intelligence Test (SIT)		Language	No difference in rate of growth of achievement by IQ; that is, those entering with lower IQ, end with lower IQ.	F–test

Summary of Effects on Student Achievement for Direct Instruction

Source

Research
Rating

Measurement
Instrument

Group/Subgroup

Dimension
Measured

Effect

Test Statistic

Gersten, Carnine (1984)

3.4

MAT

DI vs. comparison and DI vs. 8 models² for reform at Follow-Through site vs. non-Follow-Through site

Mathematics

DI scored 48^th percentile; next best was Behavioral Analysis at 28^th percentile

Percentile

Self-concept

DI scored +18 (+ means positive self concept, -means negative), which was 3^rd.

Percentile

Coopersmith Self-Concept Inventory IARS

Sense of responsibility

DI scored +21, which was first among all 8 programs.

Percent rank

Gersten, Darch, Gleason (1988)

3.2

MAT

Third-grade students with 3 and 4 years of DI vs. local comparison group

Total Reading

Word knowledge

Reading (comprehension)

Language

Spelling

Total math

Computation

Math concepts

Problem solving

DI students with three years of DI scored higher than comparison students on language, total math, computation, math concepts, and problem solving. DI students with four years of DI scored higher than comparison students on all subtests except for spelling.

Not specified, but probability levels given

Coopersmith Self-Concept Inventory

Self-concept

Students with three years and students with four years of DI both scored more positively than comparison students (.30 and .40 effect size respectively).

Effect size

CAT

Ninth-grade follow-up group vs. local comparison group

Reading
Language
Math

Three-year DI students scored higher than comparison students in reading and language, and four-year DI students scored higher than comparison students on reading, language, and math.

t-test

Summary of Effects on Student Achievement for Direct Instruction

Source

Research
Rating

Measurement
Instrument

Group/Subgroup

Dimension
Measured

Effect

Test Statistic

Gersten, Carnine, Zoref, Cronin (1986)

3.0

CTBS

Kindergarten through second-grade DI students

Reading

Across 7 classes, students went from an average of the 18^th to an average of the 46^th percentile on CTBS, students in classes with two teachers showing highest levels of implementation (as measured by DISC) showed higher percentile scores on CTBS than students in classes with two teachers showing lowest levels of implementation, and scores on CTBS had a significant correlation with three different measures of implementation (DISC, expert judgements, and supervisor ratings).

Various

WRAT

WRAT scores correlated positively with measures of implementation (expert judgements, supervisor ratings).

Grossen, Ewing (1994)

3.2

Inventory, 4^th grade level of Scott Foresman text as pretest

DI vs. National Council of Teachers of Mathematics standards

Mathematics problem solving

No statistically significant difference between DI and NCTM students.

F–test

Algebraic word problems post-test (DI)

DI students scored significantly higher, with mean scores of 17.7 vs. 12.6.

F–test

Woodcock-Johnson applications scale posttest

No statistically significant difference between DI and NCTM students.

F–test

Iowa Test of Basic Skill, Concepts

No statistically significant difference between DI and NCTM students.

F–test

Iowa Test of Basic Skills, Problem Solving

No statistically significant difference between DI and NCTM students.

F–test

Iowa Test of Basic Skills, Operations

DI students scored significantly higher, with mean scores of 28.8 vs. 21.2.

F–test

Summary of Effects on Student Achievement for Direct Instruction

Source

Research
Rating

Measurement
Instrument

Group/Subgroup

Dimension
Measured

Effect

Test Statistic

Meyer (1984)

3.3

California Achievement Test

DI vs. comparison

Reading

DI students scored significantly higher, with grade equivalents of 9.20 vs. 8.21.

t-test

California Achievement Test

Mathematics

DI students scored significantly higher, with grade equivalents of 8.59 vs. 7.95.

t-test

Graduated high school

High school success

DI students graduated at a significantly higher rate than the control (59.5% vs. 37.6%).

t-test

Held back at least one grade

DI students were held back at a significantly lower rate than the control (21.4% to 32.6%).

t-test

Dropped out

DI students dropped out at a significantly lower rate than the control (27.7% to 46%).

t-test

Applied to college

DI students applied to college at a significantly higher rate than the control (34% to 18.5%).

t-test

Accepted to college

DI students were accepted to college at a significantly higher rate than the control (34% to 17%).

t-test

Meyer, Gersten, Gutkin (1983)

3.3

MAT

Students with 4 years of DI vs. local and "pooled" comparison

Math

Reading

Language

Two cohorts tested. For cohort 1, DI students scored significantly higher than pooled comparison group on math, reading, and language, and scored higher than the local comparison group on language only. For cohort 2, DI students scored significantly higher than both the pooled comparison and local comparison groups in math and language. Although no specific comparisons were made, scores for DI students were relatively consistent on both the MAT and SAT across a 9-year period.

ANCOVA

Tarver, Jung (1995)

3.2

Comprehensive Test of Basic Skills—Mathematics

DI vs. discovery learning mathematics curriculum (MTW/CGI)

Mathematics

DI students scored significantly higher on math computation and total mathematics, no difference on concepts and applications at end of first grade; DI students scored significantly higher on all 3 at end of 2nd grade.

F – test

Summary of Effects on Student Achievement for Direct Instruction

Source

Research
Rating

Measurement
Instrument

Group/Subgroup

Dimension
Measured

Effect

Test Statistic

Varela-Russo, Blasik, Ligas. (1997)

3.3

Stanford Achievement Test (SAT8) Total Reading

Alliance of Quality Schools (DI) vs. district-wide comparison

Reading

DI students had lower mean scores but showed a greater gain over one year than comparison in 3rd and 4th grades, but had lower scores and lost more in 5th grade

None

SAT8 Total Math

Mathematics

DI students had lower mean scores but showed a greater gain over one year than comparison in 3rd and 4th grades, but had lower scores and lost more in 5th grade.

None

Florida Writes

Writing

32% of DI schools improved scores; 21% outscored district average.

None

Wellington (1994)

3.2

Teacher designed

DI vs. comparison in 8 elementary schools, 1st and 4th grades

Mathematics

DI outscored comparison in one school 26.21% to 22.18%, and comparison outscored DI in one school 31.06% to 28.94% in 1st grade; in 4th grade, DI outscored comparison at 5 of 6 schools.

t–test

Summary of Effects on Student Achievement for
Expeditionary Learning Outward Bound

Source	Research Rating	Measurement Instrument	Group/Subgroup	Dimension Measured	Effect	Test Statistic
Academy for Educational Development (1995)	3.2	Metropolitan Achievement Test/ITBS	ELOB school sites; no comparison	Reading Math	Increase in scores from 1993-1995 for grades 5 and 6 (Hernandez); grade 7 (King); grades 7 and 8 (SPCELC); grade 6 (Bryant and Lincoln). Increased scores for grades 5 and 6 (Hernandez); grade 8 (SPCELC); grade 6 (Bryant and Table Mound). Decreased scores for grade 7 (King and SPCELC).	na na Not given
Expeditionary Learning (1997)	3.1	Maine Educational Assessment	King Middle School vs. state average	Reading	King increased 45 points vs. statewide avg. increase of 5 points
				Writing	King increased 55 points vs. statewide avg. increase of 0 points
				Mathematics	King increased 65 points vs. statewide avg. increase of 25 points
				Science	King increased 80 points vs. statewide avg. increase of 10 points
		Iowa Test of Basic Skills	Clairemont Elementary School, fifth grade vs. district and state average	Reading	ELOB students scored at 7.6 grade equivalent
		Iowa Test of Basic Skills		Mathematics	ELOB students scored at 8.1 grade equivalent
		Georgia Curriculum Based Assessment Test		Reading	ELOB students scored at 99^th percentile 84% of ELOB achieved at the highest standard (vs. 64% in district and 61% statewide)	Percentile
				Mathematics	ELOB students scored at 95^th percentile 84% of ELOB achieved at the highest standard (vs. 46% in district and 32% statewide)	Percentile
				Science	ELOB students scored at 98^th percentile 42% of ELOB achieved at the highest standard (vs. 18% in district and 12% statewide)	Percentile
				Social studies	ELOB students scored at 95^th percentile 32% of ELOB achieved at the highest standard (vs. 18% in district and 16% statewide)	Percentile

Summary of Effects on Student Achievement for Expeditionary
Learning Outward Bound

Source	Research Rating	Measurement Instrument	Group/Subgroup	Dimension Measured	Effect	Test Statistic
Expeditionary Learning (1997), continued		Stanford 9	Rafael Hernandez School, Boston	Mathematics	11^th out of 76 elementary schools in district in percent of fifth graders above basic proficiency.
		Stanford 9	Rafael Hernandez School, Boston	Reading	17^th out of 76 elementary schools in district in percent of fifth graders above basic proficiency.
		Iowa Test of Basic Skills	Rocky Mountain School of Expeditionary Learning, Denver	Reading	Scores increased in 7 of 9 grades.
			Rocky Mountain School of Expeditionary Learning, Denver	Language	Scores increased in 6 of 9 grades.
			Lincoln Elementary School, Dubuque, IA	Not specified	Scores of fourth graders increased from 43^rd percentile to 80^th percentile.
			Table Mound Elementary School, Dubuque, IA	Not specified	Scores of fourth graders increased from 39^th percentile to 80^th percentile.
			Bryant Elementary School, Dubuque, IA	Reading	Percent of students at 75^th percentile increased from 24% to 43%.
			Bryant Elementary School, Dubuque, IA	Mathematics	Percent of students at 75^th percentile increased from 41% to 48%.
			School for the Physical City, New York	Reading	Ranked 29^th of 226 junior high schools; 75% of students at grade level vs. 47% districtwide.

Summary of Effects on Student Achievement for Expeditionary
Learning Outward Bound

Source	Research Rating	Measurement Instrument	Group/Subgroup	Dimension Measured	Effect	Test Statistic
Farrell, Leibowitz (1998)	3.0	Iowa Test of Basic Skills	ELOB (Table Mound, Lincoln, Bryant Elementaries) Cohort 1 vs. district	Reading	Scores increased from 12^th to 42^nd percentile at Table Mound, from 6^th to 71^st at Lincoln,and 27^th to 82^nd at Bryant vs. a decrease from 56^th to 52^nd districtwide.	Percentile
				Math computation	Scores increased from 4^th to 91^st percentile at Table Mound, from 10^th to 49^th at Lincoln, and 29^th to 99^th at Bryant vs. from 52^nd to 58^th districtwide.
				Composite	Scores increased from 4^th to 51^st percentile at Table Mound, from 4^th to 54^th at Lincoln, and 29th to 85th at Bryant vs. from 47^th to 52^nd districtwide.
			ELOB (Table Mound, Lincoln, Bryant Elementaries) Cohort 2 vs. district	Reading	Scores increased from 30^th to 58^th percentile at Table Mound, from 33^rd to 62^nd at Lincoln, and 77^th to 90^th at Bryant vs. no change at 52^nd districtwide.
				Math	Scores increased from 46^th to 79^th percentile at Table Mound, from 31^st to 70^th at Lincoln, and 89^th to 90^th at Bryant vs. from 55^th to 56^th districtwide.
				Math computation	Scores increased from 26^th to 85^th percentile at Table Mound, from 7^th to 35^th at Lincoln, and 47^th to 63^rd at Bryant vs. from 45^th to 51^st districtwide.
				Composite	Scores increased from 28^th to 58^th percentile at Table Mound, from 30^th to 43^rd at Lincoln, and 62^nd to 86^th at Bryant vs. from 48^th to 50^th districtwide.

NEXT PAGE OF APPENDIX C

APPENDIX C SUMMARY OF FINDINGS ON STUDENT ACHIEVEMENT EFFECTS

APPENDIX C
SUMMARY OF FINDINGS ON
STUDENT ACHIEVEMENT EFFECTS